Sentence level pitch accent for 標準語

Question

I was wondering if anybody had any references (preferably freely available) or insights that explain 標準語 (or should I say 関東語) sentence-level pitch accent.

I already have a dictionary which describes pitch accents on individual words, but there seem to be cases where the rules change when the words are used in sentences.

To illustrate with some examples:

• "日が" (as in "日が昇る") has no drop in pitch, but in "あの日↓は", I hear a drop after "日".
• After "あの", the pitch usually continues where it left off, but in certain combinations, like "あの↓人", I hear a drop after "あの".
• For non-single-mora words that usually have a drop between the last mora and a following particle, like "橋" ("橋↓が"), "の" seems to be an exception ("橋の上" with no drop). However, "次" seems to be an exception to this exception. I hear a drop in "次↓の".

(I could be wrong about some of these, if so please correct me.)

I hear good things about the NHK pitch accent dictionary, but does it describe pitch accent at this level as well?

I'd be thankful for any information that can bring insight on this.

Answer 1

Yes, accents change when words are combined/conjugated/etc. I'm not sure if there are any truly sentence-level phenomena, but there is definitely more going on than just "words have the same accent all the time". The NHK dictionary does include a fair bit of information about these rules.

To take your examples -- here are some answers I got from consulting the dictionary quickly. (Linguists, please forgive my half-assed terminology and use of the dreaded LH representation.)

1) This is actually more complicated than it looks! The 日 and 昇る in 日が昇る are both unaccented, and the の after an unaccented noun does not cause a drop, so the whole phrase has no accent. However, the 日 in あの日は (assuming you mean "that day" and not "that sun" IS accented. They are not actually homophones if you take pitch into account. あの is unaccented, and は coming after an accented mora = no accent, so あの日は = LHHL.

2) あの is unaccented as you say, but in combination with certain nouns, the whole phrase takes on an accent. I don't know how to account for this, but the NHK dictionary gives あの人 its own entry, with two possible patterns: LHLL or LHHH(drop) -- so maybe it is partly lexicalized, or at least the result of some rule obscure enough to justify including the result　as its own dictionary entry.

3) When の is combined with a two-syllable word with a drop after it (like 橋 as you say), the drop generally "disappears" and the result becomes unaccented. So 橋 is LH(drop) but 橋の is LHH(no drop). Why 次の should work differently is a mystery. (The dictionary lists 次の間 as an independent item with either LHLL or LHHH(no drop), so clearly it is a known phenomenon.) -- Although, it should be noted that の working this way is kind of an exception in the case of noun + particle, which usually doesn't change the accent on the noun.

The problem with the information in the NHK dictionary (I mean the general "rules" rather than the entries for individual words) is that although there is a lot of it, it is for reference rather than pedagogical purposes -- and it's not even indexed very well. So it's not easy to learn from it if what you're after is a general survey of how Japanese accent works. I imagine it works much better if you're using the book as a reference while taking a course on how to speak NHK-style.